Build a token optimization layer between AI agents and LLM providers to intercept and optimize prompts/inputs and reduce tokens consumption.

Technologies: langfuse, LiteLLM, Terraform, tiktoken.